clothing-mnist.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "#importing libraries\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import torch\n",
    "from torch import nn\n",
    "from torch import optim\n",
    "from torchvision import datasets, transforms\n",
    "\n",
    "import helper"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define a transform to normalize the data\n",
    "transform = transforms.Compose([transforms.ToTensor(),\n",
    "                                transforms.Normalize((0.5,),(0.5,))])\n",
    "\n",
    "# Download and load the training data\n",
    "trainset = datasets.FashionMNIST(\"~pytorch/F_MNIST_data/\", download=True, train=True, transform=transform)\n",
    "trainloader = torch.utils.data.DataLoader(trainset, batch_size=64, shuffle=True)\n",
    "\n",
    "# Download and load the test data\n",
    "testset = datasets.FashionMNIST(\"~pytorch/F_MNIST_data/\", download=True, train=False, transform=transform)\n",
    "testloader = torch.utils.data.DataLoader(testset, batch_size=64, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Run this to test the data loader\n",
    "images, labels = next(iter(trainloader))\n",
    "helper.imshow(images[0], normalize=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sequential(\n",
      "  (0): Linear(in_features=784, out_features=256, bias=True)\n",
      "  (1): ReLU()\n",
      "  (2): Dropout(p=0.2, inplace=False)\n",
      "  (3): Linear(in_features=256, out_features=128, bias=True)\n",
      "  (4): ReLU()\n",
      "  (5): Dropout(p=0.2, inplace=False)\n",
      "  (6): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (7): ReLU()\n",
      "  (8): Dropout(p=0.2, inplace=False)\n",
      "  (9): Linear(in_features=64, out_features=10, bias=True)\n",
      "  (10): Softmax(dim=1)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "input_layer = 784\n",
    "hidden_layers = [256, 128, 64]\n",
    "output_layer = 10\n",
    "\n",
    "model = nn.Sequential(\n",
    "                    nn.Linear(input_layer, hidden_layers[0]),\n",
    "                    nn.ReLU(),\n",
    "                    nn.Dropout(p=0.2),\n",
    "                    nn.Linear(hidden_layers[0], hidden_layers[1]),\n",
    "                    nn.ReLU(),\n",
    "                    nn.Dropout(p=0.2),\n",
    "                    nn.Linear(hidden_layers[1], hidden_layers[2]),\n",
    "                    nn.ReLU(),\n",
    "                    nn.Dropout(p=0.2),\n",
    "                    nn.Linear(hidden_layers[2], output_layer),\n",
    "                    nn.Softmax(dim=1))\n",
    "\n",
    "print(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1/30... Training Loss: -0.156... Testing Loss: -0.321... Accuracy: 0.335...\n",
      "Epoch: 2/30... Training Loss: -0.449... Testing Loss: -0.593... Accuracy: 0.630...\n",
      "Epoch: 3/30... Training Loss: -0.627... Testing Loss: -0.706... Accuracy: 0.724...\n",
      "Epoch: 4/30... Training Loss: -0.714... Testing Loss: -0.753... Accuracy: 0.764...\n",
      "Epoch: 5/30... Training Loss: -0.745... Testing Loss: -0.766... Accuracy: 0.773...\n",
      "Epoch: 6/30... Training Loss: -0.761... Testing Loss: -0.771... Accuracy: 0.775...\n",
      "Epoch: 7/30... Training Loss: -0.770... Testing Loss: -0.779... Accuracy: 0.783...\n",
      "Epoch: 8/30... Training Loss: -0.777... Testing Loss: -0.783... Accuracy: 0.788...\n",
      "Epoch: 9/30... Training Loss: -0.783... Testing Loss: -0.787... Accuracy: 0.791...\n",
      "Epoch: 10/30... Training Loss: -0.787... Testing Loss: -0.791... Accuracy: 0.793...\n",
      "Epoch: 11/30... Training Loss: -0.791... Testing Loss: -0.792... Accuracy: 0.794...\n",
      "Epoch: 12/30... Training Loss: -0.794... Testing Loss: -0.796... Accuracy: 0.798...\n",
      "Epoch: 13/30... Training Loss: -0.798... Testing Loss: -0.795... Accuracy: 0.797...\n",
      "Epoch: 14/30... Training Loss: -0.800... Testing Loss: -0.796... Accuracy: 0.798...\n",
      "Epoch: 15/30... Training Loss: -0.802... Testing Loss: -0.801... Accuracy: 0.803...\n",
      "Epoch: 16/30... Training Loss: -0.804... Testing Loss: -0.800... Accuracy: 0.802...\n",
      "Epoch: 17/30... Training Loss: -0.807... Testing Loss: -0.803... Accuracy: 0.805...\n",
      "Epoch: 18/30... Training Loss: -0.809... Testing Loss: -0.807... Accuracy: 0.810...\n",
      "Epoch: 19/30... Training Loss: -0.811... Testing Loss: -0.809... Accuracy: 0.811...\n",
      "Epoch: 20/30... Training Loss: -0.811... Testing Loss: -0.811... Accuracy: 0.812...\n",
      "Epoch: 21/30... Training Loss: -0.813... Testing Loss: -0.811... Accuracy: 0.812...\n",
      "Epoch: 22/30... Training Loss: -0.815... Testing Loss: -0.810... Accuracy: 0.811...\n",
      "Epoch: 23/30... Training Loss: -0.817... Testing Loss: -0.813... Accuracy: 0.815...\n",
      "Epoch: 24/30... Training Loss: -0.818... Testing Loss: -0.813... Accuracy: 0.816...\n",
      "Epoch: 25/30... Training Loss: -0.819... Testing Loss: -0.815... Accuracy: 0.816...\n",
      "Epoch: 26/30... Training Loss: -0.820... Testing Loss: -0.813... Accuracy: 0.815...\n",
      "Epoch: 27/30... Training Loss: -0.821... Testing Loss: -0.816... Accuracy: 0.818...\n",
      "Epoch: 28/30... Training Loss: -0.822... Testing Loss: -0.816... Accuracy: 0.818...\n",
      "Epoch: 29/30... Training Loss: -0.823... Testing Loss: -0.816... Accuracy: 0.817...\n",
      "Epoch: 30/30... Training Loss: -0.825... Testing Loss: -0.817... Accuracy: 0.818...\n"
     ]
    }
   ],
   "source": [
    "criterion = nn.NLLLoss()\n",
    "optimizer = optim.SGD(model.parameters(), lr=0.03)\n",
    "\n",
    "epochs = 30\n",
    "training_losses = []\n",
    "testing_losses = []\n",
    "\n",
    "for e in range(epochs):\n",
    "    #training the model\n",
    "    total_training_loss = 0\n",
    "    for images, labels in trainloader:\n",
    "        images = images.view(images.shape[0], -1)\n",
    "\n",
    "        optimizer.zero_grad()\n",
    "        logps = model(images)\n",
    "        loss = criterion(logps, labels)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        total_training_loss += loss.item()\n",
    "\n",
    "    else:\n",
    "        #testing the model\n",
    "        total_testing_loss = 0\n",
    "        test_correct = 0\n",
    "        with torch.no_grad():\n",
    "            model.eval() #dropout turns off\n",
    "            for images, labels in testloader:\n",
    "                images = images.view(images.shape[0], -1)\n",
    "                logps = model(images)\n",
    "                loss = criterion(logps, labels)\n",
    "                total_testing_loss += loss.item()\n",
    "\n",
    "                ps = torch.exp(model(images))\n",
    "\n",
    "                top_p, top_class = ps.topk(1, dim=1)\n",
    "\n",
    "                equals = top_class == labels.view(*top_class.shape)\n",
    "\n",
    "                test_correct += torch.mean(equals.type(torch.FloatTensor))\n",
    "                #test_correct += equals.sum().item()\n",
    "            \n",
    "        model.train() #dropout back on\n",
    "\n",
    "        training_losses.append(total_training_loss / len(trainloader))\n",
    "        testing_losses.append(total_testing_loss / len(testloader))\n",
    "\n",
    "        print(\"Epoch: {}/{}...\".format(e+1, epochs),\n",
    "        \"Training Loss: {:.3f}...\".format(training_losses[-1]),\n",
    "        \"Testing Loss: {:.3f}...\".format(testing_losses[-1]),\n",
    "        \"Accuracy: {:.3f}...\".format(test_correct/len(testloader)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fd0a9c22f40>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(training_losses, label=\"Training loss\")\n",
    "plt.plot(testing_losses, label=\"Testing loss\")\n",
    "plt.legend(frameon=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(7)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x648 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "images, labels = next(iter(testloader))\n",
    "images.resize_(images.shape[0], 1, 784)\n",
    "ps = model.forward(images[0,:])\n",
    "\n",
    "helper.view_classify(images[0].view(1,28,28), ps)\n",
    "print(labels[0])"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "58e01130d00a3632b0fb78b77be37655486778a777e7091d9e1671094e6b3d94"
  },
  "kernelspec": {
   "display_name": "Python 3.9.5 64-bit (conda)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}